This document is meant to be provide some quick reference materials for gettings started with the basics of working with scripts against a slurm based computational cluster.
The 5 scripts below should all be created or downloaded into the same folder and submit commands run from that base directory.
| #!/usr/bin/env bash | |
| # This will submit a script via sbatch and capture the job id as a variable for use in bash | |
| job1taskid=`sbatch --parsable ./01job.sh` | |
| # This then submits a job that waits for the first job to finish before running | |
| sbatch --dependency=$job1taskid ./02job.sh |
| #!/usr/bin/env bash | |
| ## This is just a small test job to print some output and sleep so we can demostrate dependencies | |
| #SBATCH -J 'initaljob' | |
| #SBATCH -o slurm-%j.out | |
| #SBATCH -p all | |
| #SBATCH -t 5 | |
| echo "INITAL TASK STARTS" | |
| sleep 60 |
| #!/usr/bin/env bash | |
| ## This job will be scheduled at the same time as 01job.sh but will wait for that to finish | |
| ## before running. | |
| #SBATCH -J 'secondjob' | |
| #SBATCH -o slurm-%j.out | |
| #SBATCH -p all | |
| #SBATCH -t 5 | |
| echo "second task ran" |
| #!/usr/bin/env bash | |
| #this script dumps some basic information to an output file when submitted via sbatch | |
| #name the job nodeinfo and place it's output in a file named slurm-<jobid>.out | |
| #set partition to 'all' this isn't strictly necessary but it's good practice | |
| #set time to 5 minutes so jobs get killed if something weird happens | |
| #SBATCH -J 'nodeinfo' | |
| #SBATCH -o slurm-%A_%a.out | |
| #SBATCH -p all | |
| #SBATCH -t 5 | |
| #SBATCH -c 2 | |
| # clean up environment so submission environment doesn't impact the job | |
| module purge | |
| module load suite2p/0.10.2 | |
| # Dump test info about the specific host the job is being run on | |
| echo "In the directory: `pwd` " | |
| echo "As the user: `whoami` " | |
| echo "on host: `hostname` " | |
| # Dump the job array information if it's available | |
| echo "Array ID: $SLURM_ARRAY_JOB_ID" | |
| echo "Array Index: $SLURM_ARRAY_TASK_ID" | |
| # test to make sure you are getting all the cpu cores you asked for | |
| echo "With access to cpu id(s): " | |
| cat /proc/$$/status | grep Cpus_allowed_list | |
| # run the small python demo script used to demonstrate access env variables inside a python script | |
| ./pyenvprint.py |
| #!/usr/bin/env python | |
| # this brings in tools to navigate the OS directly | |
| import os | |
| # this captures the array task id if it is set as an environmental variable | |
| taskid=None | |
| taskid=os.getenv('SLURM_ARRAY_TASK_ID') | |
| # check if we're running an array job or not to decide what we'll be outputing into the slurm output. | |
| if taskid: | |
| print('array task id: ',taskid) | |
| else: | |
| print(os.environ) |